Interannual climate variability and population density thresholds can have a substantial impact on simulated tree species' migration

Assessments of future tree species' distributions should account for time lags in the adaptation of their external range limits to climatic changes. In simulation experiments it is therefore necessary to capture processes that influence such time lags, in particular tree species' migration. We hypothesise that directional processes such as migration are sensitive to the exact sequence of simulated climate influences, and that the uncertainty associated with a given interannual climate variability has to be accounted for when simulating migration explicitly. In this paper we used the intermediate-complexity multi-species model TreeMig to examine whether different realisations of future climate influences with the same temporal mean and the same interannual variability cause fundamental differences in simulated migration. We assume that the impact of interannual climate variability becomes most apparent in situations which critically influence regeneration and survival. Such situations arise, for example, when species' sensitivities to climate, competition and spatial fragmentation interact. We therefore developed an illustrative and realistic simulation setup representing this situation. We simulated the northwards migration of the sub-Mediterranean tree species Ostrya carpinifolia Scop. (European Hop Hornbeam) through the highly fragmented and climatically heterogeneous landscape of the Swiss Alps. Situations critically influencing regeneration and survival can lead to low species' abundances. Before investigating effects of interannual climate variability, we therefore tested whether the continuous representation of species' cell populations in TreeMig, which allows for infinitesimal population densities, can have side effects on simulated migration. Specifically, we tested for effects of minimum density thresholds, i.e. thresholds below which a species is treated as absent. We found that small threshold s in the magnitude of one individual per km(2) cell have a considerable impact on simulated migration, and can even impede migration in situations critical for regeneration and survival. To test for effects of interannual climate variability, we compared simulation results from multiple repetitions driven by different annual climate time series generated stochastically from the same probability distribution. Results from these repetitions were additionally compared to results from simulations driven by cyclically repeated climate and steadily applied mean climate, respectively. These comparisons were conducted for different species parameter sets within the plausible parameter range of O. carpinifolia to account for potential interactions between species' sensitivities and the environment. Simulated tree species' migration was highly dependent on the species parameters applied and markedly influenced by interannual climate variability. Notable divergence in species' spread resulted amongst multiple realisations of annual climate time series stochastically sampled from the same probability distribution. We conclude that uncertainty associated with interannual climate variability has to be accounted for. Single realisations can be insufficient and mean value simulations as well as averages of output results can be too simplistic to reflect possible outcomes of tree species' migration. (C) 2013 Elsevier B.V. All rights reserved.